Ep 11: Awakening the Model — Wiring LLM Credentials & Basic Chat Model Nodes

From Automation to Intelligence

Episodes 1-10 had deterministic workflows — input A always produces output B. Now we introduce probabilistic LLMs, granting workflows the ability to understand language, reason, and generate content.

graph TB
    subgraph "Ep 01-10: Deterministic"
        D1[Input] --> D2[Fixed Rules] --> D3[Certain Output]
    end
    subgraph "Ep 11-30: Intelligent"
        I1[Input] --> I2[🤖 LLM Reasoning] --> I3[Dynamic Output]
        I2 -->|"May call tools"| I4[External APIs]
        I4 --> I2
    end
    style I2 fill:#ff6d5b,stroke:#e55a4e,color:#fff

1. n8n AI Node Hierarchy

graph TB
    subgraph "AI Node Architecture"
        Top[🤖 AI Agent Node
Top layer: autonomous decisions]
        Top --> LLM[🧠 Chat Model
Engine: OpenAI / Claude / Gemini]
        Top --> Memory[💾 Memory Node
Conversation history]
        Top --> Tools[🔧 Tool Nodes
Callable external abilities]
    end
    style Top fill:#ff6d5b,stroke:#e55a4e,color:#fff
    style LLM fill:#8b5cf6,stroke:#7c3aed,color:#fff

⚠️ Chat Model nodes cannot be directly wired into workflow connections. They must be embedded inside an AI Agent or LLM Chain node as sub-nodes.

2. Credential Setup

# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
# Getting your OpenAI API Key
# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
# 1. Visit https://platform.openai.com/api-keys
# 2. Click "Create new secret key"
# 3. Name it "n8n-production"
# 4. Copy the key (shown only once!): sk-proj-xxxxxxxx
# 5. Paste into n8n Credentials manager
#
# ⚠️ NEVER put API Keys in workflow expressions!
# Always use n8n's Credentials manager (AES-256 encrypted)

3. Model Comparison

4. Basic LLM Chain

// ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
// Basic LLM Chain configuration
// ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

// System Prompt:
// "You are a product copywriter. Generate a 50-word marketing blurb
//  for the given product. Be upbeat and include one emoji."

// User Message (expression):
// "Product: {{ $json.productName }}, Features: {{ $json.features }}"

// Chat Model: gpt-4o-mini, Temperature: 0.7, Max Tokens: 200

5. Temperature Guide

Next Episode

In Ep 12, we upgrade from basic LLM Chain to a full AI Agent — connecting Chat Trigger to build a multi-turn conversational bot.